On average, the cases and comparisons in the study were 46 years of age and had occupied the building for 6 years (Table 1). More than half of them were white (69%), never-smokers (61%), and female (59%). There were fewer white employees and never-smokers but more females in the case groups than in the comparison group. The proportion of current smokers was lowest (6.1%) in the postoccupancy asthma cases.

We collected 338 floor and 327 chair dust samples from 323 employees’ workstations among the 352 case and comparison group employees. We could not locate workstations for 29 participants. For those who had multiple samples because of changes in their workstations between September 2001 and April 2002, we assigned measurements of microbial agents from the workstation they occupied during the 2001 questionnaire survey. Because of the limited amount of dust collected for some samples, we prioritized sample analysis by endotoxin, culturable fungi, ergosterol, and allergens. We recovered a total of 67 fungal species from floor dust samples and 69 species from chair dust samples. In addition, unidentifiable species of Penicillium, yeasts (Rhodotorula and Sporobolomyces), and nonsporulating fungi were cultured. The GM of total culturable fungi was 7,700 colony-forming units (CFU) per gram in floor dust, which was significantly (p < 0.005) lower than that (11,000 CFU/g) in chair dust (Table 2). In the floor dust, on average, 57% of total fungal colonies were identified as hydrophilic fungi and 19% as mesophilic. In the chair dust, on average, 45% of total fungal colonies were identified as hydrophilic and 28% as mesophilic. Eighty-seven percent of the hydrophilic fungi in floor dust and 74% of those in chair dust were yeasts. GMs of the ergosterol (0.5 ng/mg) and endotoxin (10.9 EU/mg) levels in floor dust were significantly (p-values < 0.002) higher than those in chair dust (0.4 ng/mg and 2.1 EU/mg, respectively). The levels of cat (GM, 2.5 μg/g) and dog (2.1 μg/g) allergens were significantly (p-values < 0.0001) lower in floor dust than in chair dust (GMs, 12.5 μg/g and 5.7 μg/g, respectively) (Table 2).

Aureobasidium pullulans and Epicoccum nigrum were the most prevalent of the mesophilic fungi (Table 3). Yeasts and Phoma herbarum were the most prevalent hydrophilic fungi recovered, and their median concentrations (2,400 CFU/g in floor dust and 2,340 CFU/g in chair dust for yeasts; 2,850 CFU/g in floor dust and 3,600 CFU/g in chair dust for Phoma herbarum) were among the highest found for the mesophilic and hydrophilic fungi. Rhodotorula was the predominant genus of yeast identified from floor (49.4%) and chair (46.0%) dust samples. Among those that were not classified as having an A_w of ≥ 0.8, Penicillium chrysogenum (12.5%) and Pithomyces chartarum (10.4%) were the most frequently found species in floor dust samples, with Pithomyces chartarum (23.3%) and Phoma glomerata (11.7%) most frequently in chair dust samples. Penicillium chrysogenum and Aspergillus niger were the most prevalent Penicillium/Aspergillus species identified in both floor and chair dust. Nonidentifiable Penicillium species were found in 17.1% and 13.5% of the floor and chair dust samples, respectively. We found Stachybotrys chartarum in four floor samples and five chair samples.

The levels of ergosterol per square meter or chair were significantly (p-values < 0.0001) but poorly correlated with total fungi, fungi requiring A_w ≥ 0.8, hydrophilic fungi, and yeasts both in floor and chair dust samples (r = 0.15–0.27). Ergosterol had a higher correlation with endotoxin in floor dust (r = 0.47) and chair dust (r = 0.34). Ergosterol and culturable fungi were also significantly but weakly correlated with cat and dog allergens in both floor and chair dust (r = 0.12–0.31). Endotoxin had a higher correlation with cat (r = 0.39) and dog allergens (r = 0.36) in chair dust than in floor dust (r = 0.25 for cat and 0.21 for dog allergens).

The number of subjects in each of the statistical models varied, depending on the number of subjects with information missing in any of the demographic variables or the exposure variables. We found significant linear exposure–response relationships between various microbial measurements (total fungi, fungi requiring A_w ≥ 0.8, hydrophilic fungi, ergosterol, and endotoxin) in dust and health outcomes (respiratory cases, epi-asthma cases, and postoccupancy asthma cases) in the single environmental variable models. We found no significant associations (at α = 0.05) among health outcomes and fungi that were not classified as having an A_w ≥ 0.8, or fungi that were not classified as hydrophilic (Table 4). In GAMs, test p-values of these environmental variables for a null hypothesis of linearity were > 0.05, which indicates no evidence of nonlinear relationships between exposure and health outcomes. The associations between health outcomes and fungi were mostly driven by exposure to fungi requiring A_w ≥ 0.8, and specifically hydrophilic fungi in both floor and chair dust.

In single environmental variable models, we found > 65% increases in the odds of being a respiratory case for increasing IQR exposure [IQR-OR = 1.66–1.73; p-values < 0.05] for total fungi, fungi requiring A_w ≥ 0.8, and hydrophilic fungi in floor dust (IQR for each of the microbial measurements for all dust samples analyzed are presented in Table 2). Yeasts in chair dust and non-yeast hydrophilic fungi in floor dust were significantly associated with increases in the odds of being a respiratory case of 46% and 50%, respectively (Table 4). Ergosterol and endotoxin in floor dust were associated with significantly increased odds of being a respiratory case (IQR-OR = 1.56 and 1.60, respectively). Cat and dog allergens in floor dust were not significantly associated with respiratory cases at α = 0.05. In general, the odds of being a respiratory case for chair dust were lower than those for floor dust, except for the yeasts.

When compared to respiratory case outcome models, we found slightly larger magnitudes of IQR-ORs (range, 1.46–1.80) for epi-asthma cases associated with total culturable fungi, fungi requiring A_w ≥ 0.8, hydrophilic fungi, and ergosterol in both floor and chair dust (Table 4). When we examined the associations of those who reported physician-diagnosed HP or having two or more HP-like symptoms with microbial exposures, we found similar magnitudes of ORs for fungi but stronger associations with ergosterol (IQR-OR = 1.93) and endotoxin (IQR-OR = 1.80) in floor dust, and yeasts in chair dust (IQR-OR = 1.65) (data not shown).

In the models with postoccupancy asthma cases as an outcome, the associations with fungi requiring A_w ≥ 0.8, hydrophilic fungi, and yeasts in both floor and chair dust were stronger than those in the models with respiratory and epi-asthma cases as outcomes (Table 4). Of all the environmental variables, hydrophilic fungi in floor dust (IQR-OR = 2.19) were most strongly associated with post-occupancy asthma cases. Increased exposure to yeasts in floor dust and hydrophilic fungi without yeasts in chair dust by IQR significantly increased the odds of being a postoccupancy asthma case by 77% and 44%, respectively.

When we ran models with culturable fungi, ergosterol, endotoxin, and demographic variables simultaneously, the ORs of the respiratory and epi-asthma cases for total fungi and hydrophilic fungi in both floor and chair dust were slightly smaller (IQR-OR = 1.46–1.62 for floor dust and IQR-OR = 1.36–1.57 for chair dust) but generally remained significant at α = 0.05. The exception was the modeling of respiratory cases with total chair fungi (IQR-OR = 1.36; p = 0.06) adjusted for ergosterol, endotoxin, and demographic variables (Table 5). In the total fungi models of the multiple environmental variable models, the magnitude of the IQR-ORs for the respiratory and epi-asthma cases associated with total culturable fungi was smaller than that for hydrophilic fungi in the hydrophilic fungi models. Exposure to hydrophilic fungi in floor and chair dust were associated with about a 2-fold increase in the odds of being a post-occupancy asthma case (IQR-OR = 2.09 for floor dust and IQR-OR =1.79 for chair dust) (Table 5). GAMs with nonlinear spline smoothing functions did not provide evidence of nonlinear relationships; that is, the log odds (logit) of respiratory illnesses in all of these multiple environmental variable models increased linearly with increase of exposure. A sensitivity analysis was performed by rerunning all statistical models after assigning floor-specific mean values (new models) of microbial measurements to the respiratory cases and comparisons with no individual measurements. For each exposure variable we calculated the ratio (expressed as percentage) of the OR for the new model to the original model. The results showed that the models were not substantially sensitive because the ORs from the new models ranged from 84.8% to 102.8% of the original ORs presented in Tables 4 and 5.

Among employees of a water-damaged office building, we found linear associations between respiratory illnesses and the levels of fungi, which were largely explained by hydrophilic (water-loving or tertiary) fungi including yeasts. These linear exposure–response relationships based on individual exposure measurements extend our previous findings, which were based on exposure assigned as floor-specific mean values of fungal measurements (Park et al. 2006). The present study indicates that exposure assessment using individual dust samples has an advantage over assigning exposure using floor-specific mean values, because we found associations between physician-diagnosed postoccupancy-onset asthma and fungal exposure that were not demonstrated in the previous study (Park et al. 2006). The enhanced findings are likely to be explained by minimization of exposure misclassification by using individual samples in exposure assessment. Of the three respiratory health outcomes studied, asthma diagnosed after building occupancy had the strongest association with the levels of hydrophilic fungi in dust. This finding implies that a more specific (less sensitive) definition of outcome based on physician diagnosis, which likely represents more severe disease, probably minimized health outcome misclassification in relation to building exposures. We previously reported that the incidence of adult-onset asthma in this population was 7.5 times higher after building occupancy (Cox-Ganser et al. 2005). Taken together, these findings are consistent with the involvement of building-related fungal exposure in the causal chain of adult-onset asthma, although we cannot rule out that fungi—specifically hydrophilic species—may be simply markers of other causative agents in damp environments.

Among hydrophilic fungi, yeasts in both floor and chair dust played an important role in associations we found for increased odds of respiratory illnesses. Yeasts have been reported to be among the most abundant fungi found in indoor air (Cheong et al. 2004; Rantio-Lehtimaki 1988) and in house dust (Flannigan et al. 1993; Verhoeff et al. 1994). Similarly, in the present study, the Rhodotorula genus of yeasts was one of the most prevalent and abundant fungi recovered in floor and chair dust. Rhodotorula species have been shown to be implicated in IgE-mediated allergy responses (Day and Ellis 2001), as well as being potential causative agents of HP in case studies (Hodges et al. 1974; Siersted and Gravesen 1993). Our study also showed that yeasts in chair dust were significantly associated with an epidemiologic definition of HP (those who reported physician-diagnosed HP or having two or more HP-like symptoms). In occupational environments, such as bakeries, breweries, and distilleries, the yeast Saccharomyces cerevisiae is a major allergen source for allergic diseases (Day and Ellis 2001). Without yeasts, the hydrophilic fungi as a group (Phoma herbarum, Chaetomium globosum, Mucor plumbeus, Rhizopus stolonifer, and Stachybotrys chartarum) were also strongly associated with odds of respiratory illnesses. In these models, we assigned a value of LOD/2 (200 CFU/g) to a large proportion of samples < LOD (Table 2), and these assigned values were multiplied by the amount of dust collected for each subject to obtain CFUs per square meter or per chair. This assignment might have produced nondifferential misclassification in exposure, resulting in the underestimation of ORs. Even with this potential misclassification, we still found significant associations between hydrophilic fungi without yeast and respiratory illnesses. Phoma spp., Mucor spp., and Rhizopus spp. have been implicated in IgE-mediated allergy (Day and Ellis 2001). However, we are not aware of epidemiologic studies demonstrating increased risk of building-related asthma or other respiratory illnesses associated with yeasts and other hydrophilic fungi in floor and chair dusts in water-damaged nonindustrial buildings.

We found poor correlations (r < 0.3) between ergosterol levels and culturable fungi in more than 300 dust samples, although Saraf et al. (1997) reported a higher correlation (r = 0.65) in 17 house dust samples. This is not a completely unexpected result for the following reasons. First, ergosterol is found in mycelia and fungal fragments, as well as in intact spores (Miller 2001). Second, ergosterol can be detected in both viable and nonviable spores. Third, culturable fungi represent only a small portion of the viable spores that can grow on the selected media (Dales et al. 1999; Saraf et al. 1997). Finally, the proportion of viable and nonviable spores may differ across the samples, and the proportion of the viable spores that can be cultured may also differ. Hence, ergosterol has the potential to measure fungal biomass more accurately than the culture technique. Because health effects such as allergy and inflammation do not rely on viability of fungal contaminants, measuring ergosterol to estimate total fungal biomass in exposure assessment is warranted (Nielsen and Madsen 2000; Pasanen et al. 1999).

A few research groups have used ergosterol measurements for exposure assessment in epi-demiologic studies, but the results have been inconsistent. In a cross-sectional analysis of the European Community Respiratory Health Survey (ECRHS) subcohort followed up in 1996 (n = 485), Dharmage et al. (2001) found a significant association of ergosterol levels in bedroom dust with sensitization to fungi and having wheezed in the last year. They also performed a longitudinal analysis on repeated measurements of wheeze and ergosterol in bedroom dust in 1996 and 1998 on the same ECRHS subcohort. They found a statistically significant interaction effect in which the effect of increasing ergosterol on the chance of remission of wheeze depended on the initial levels of ergosterol in 1996 (Matheson et al. 2005). From the same ECRHS subcohort, 35 young asthmatic adults sensitized to fungi were followed over four seasons. No association was found between either culturable fungi or ergosterol levels in bedroom floor dust and peak flow variability. However, Dharmage et al. (2002) discussed that this lack of association could have been partly explained by misclassification of exposure. In a study of children in Canada, Dales (1998) reported a significant association between living in fungal-contaminated homes with higher airborne ergosterol levels and an increased number of CD3+ T cells expressing CD45RO. In a later study, Dales et al. (1999) examined the association of airborne ergosterol levels in bedrooms with respiratory symptoms and nocturnal cough among elementary school children, but they did not find associations. In that study (Dales et al. 1999), airborne ergosterol levels were estimated from less-than-a-day sampling. This might have misclassified children’s exposure in relation to the 1- and 12-month time periods covered by the symptom questionnaire. Mendell et al. (2002) measured airborne ergosterol in a double-blind cross-over study evaluating the effect of replacement of filters on occupants’ symptoms and indoor particles in an office building; they found ergosterol levels < LOD for seven of eight air samples.

In contrast, we found that 230- or 260-ng increases in ergosterol levels per square meter of floor or per chair, respectively, elevated the odds of respiratory illnesses (especially epi-asthma cases) by 46–55% in models adjusted for demographics, culturable fungi, and endotoxin. It is not known whether ergosterol can directly induce respiratory health effects or if it is a surrogate measure of exposure to fungi or of another unmeasured exposure related to dampness. However, our finding of associations between ergosterol and health outcomes, independent of culturable fungi, suggests that measuring both ergosterol and culturable fungi may be important to fully understand health effects associated with fungal exposure in epidemiologic studies.

In the building we studied, which had a long history of water damage, the first major construction activity related to water incursion began in 2000, with repair of roof copings and brick caulking. From 2000 to 2002 cubicle partitions and carpets were cleaned, wetted carpet and stained wallboard replaced, wall-paper and underlying mold removed from bathrooms, upgrades to the air handling system made, and windows caulked. We found that the levels of culturable fungi in dust sampled in 2002, 7 months after the 2001 questionnaire survey, were low compared with those in other studies of office buildings (Chao et al. 2001), school buildings (Ebbehoj et al. 2005), and residential buildings (Hicks et al. 2005) with no apparent water damage. Before our study, the historical levels of fungi were higher. For example, three consultant reports on 20 surface dusts from 2000 to 2001 showed a range of 4,700–7,800,000 CFU/g as compared to our current study range of 276–1,200,000 CFU/g in floor dust (Occupational Risk Control Services, unpublished data). Furthermore, fungal contamination was found in the walls. We surmise that the relative differences in occupant exposure and fungal profile in the dust at individual workstations might have remained even though the remediation action changed the absolute levels of microbial contaminants. This would explain the association between the fungal exposure and health effects even at the low absolute levels of fungi in the dust.

The fungal profile in our dust samples was predominantly hydrophilic and mesophilic fungi. At the time of our study, the carpet and chairs generally showed low A_w (0.18–0.8, with a mean of 0.5). Historical reports indicated that the building had extensive water damage in the past, and our fungal profile analyses support that there must have been wet conditions. Hydrophilic fungi are not likely to become predominant unless wet conditions persist for an extended time (Flannigan and Miller 2001). In a study of houses without water damage, water-indicator fungi (Chaetomium spp., Ulocladium spp., and Stachybotrys spp.) were largely absent from air and dust samples (Horner 2004). Furthermore, because the spores of the four most dominant fungi we found in both floor and chair dusts (yeasts, Aureobasidium pullulans, Alternaria alternata, and Epicoccum nigrum) have long survival times (Flannigan and Miller 2001), they can remain in relatively dry conditions as indicators of past dampness.

Several different fungal genera and species have been used as indicators of water-damaged indoor environments (Burge and Otten 1999; Flannigan and Miller 2001). We know of no source that lists the water activities for all fungi. Furthermore, minimum and optimum A_w for growth of individual fungal species can differ depending on environmental conditions, such as temperature and nutrient availability (Burge and Otten 1999). We categorized mesophilic and hydrophilic fungi based on three reports (Burge and Otten 1999; Flannigan and Miller 2001; Grant et al. 1989), and thus we may have some misclassification. However, because most of the fungi with prevalence of > 10% were classifiable based on those reports, the misclassification in A_w categorization is not likely to change our findings in the study.

In conclusion, we showed that among employees in a building with a long history of water damage, respiratory symptoms and post-occupancy asthma were strongly associated with fungi in a linear exposure–response manner, especially the levels of hydrophilic fungi (including yeasts) in dust. These findings extend the conclusions of insufficient evidence for the development of asthma in relation to the presence of mold or other agents in damp indoor environments reported by the Institute of Medicine (2004). Because the markers (total culturable fungi, hydrophilic fungi, and ergosterol) of potential mold exposure were associated with health outcomes, we suggest that further research to understand respiratory health effects in water-damaged indoor environments include measurements of both ergosterol and speciated culturable fungi in dust.